Article
Effect of the Network Structure and Programming Temperature on the Shape-Memory Response of Thiol-Epoxy “Click” Systems Alberto Belmonte 1 , Dailyn Guzmán 2 , Xavier Fernández-Francos 3 and Silvia De la Flor 1, * Received: 28 July 2015 ; Accepted: 21 October 2015 ; Published: 26 October 2015 Academic Editor: Wei Min Huang 1 2 3
*
Department of Mechanical Engineering, Universitat Rovira i Virgili, Av. Països Catalans 26, 43007 Tarragona, Spain;
[email protected] Department of Analytical and Organic Chemistry, Universitat Rovira i Virgili, C/Marcel¨ lí Domingo s/n, 43007 Tarragona, Spain;
[email protected] Thermodynamics Laboratory, ETSEIB, Universitat Politècnica de Catalunya, Av. Diagonal 647, 08028 Barcelona, Spain;
[email protected] Correspondence:
[email protected]; Tel.: +34-97-755-8839; Fax: +34-97-755-9602
Abstract: This paper presents a new methodology to develop “thiol-epoxy” shape-memory polymers (SMPs) with enhanced mechanical properties in a simple and efficient manner via “click” chemistry by using thermal latent initiators. The shape-memory response (SMR), defined by the mechanical capabilities of the SMP (high ultimate strength and strain), the shape-fixation and the recovery of the original shape (shape-recovery), was analyzed on thiol-epoxy systems by varying the network structure and programming temperature. The glass transition temperature (Tg ) and crosslinking density were modified using 3- or 4- functional thiol curing agents and different amounts of a rigid triglycidyl isocyanurate compound. The relationship between the thermo-mechanical properties, network structure and the SMR was evidenced by means of qualitative and quantitative analysis. The influence of the programming temperature (Tprog ) on the SMR was also analyzed in detail. The results demonstrate the possibility of tailoring SMPs with enhanced mechanical capabilities and excellent SMR, and intend to provide a better insight into the relationship between the network structure properties, programming temperature and the SMR of unconstrained (stress-free) systems; thus, making it easier to decide between different SMP and to define the operative parameters in the useful life. Keywords: epoxy; thiol; shape-memory polymer; click chemistry
1. Introduction Shape-memory polymers (SMPs) are materials that can change from a temporary shape to their original shape through an external stimulus. The original shape is given by the curing process and it is permanent while the temporary shape is imposed by a programming process, which depends on the physical nature of the shape-memory polymer (SMP) [1–4]. The most common type are the thermo-responsive SMPs [5]. The efficiency of a SMP is defined by the shape-memory response (SMR), which includes the mechanical capabilities (high ultimate strength and strain), good fixation and retention of the temporary shape and well-controlled shape-recovery process. Shape-memory thermosets are recently the focus of several researchers due to the higher strength, larger durability, better retention of the temporary shape and faster recovery of the original shape than those SMPs based on thermoplastics. Specifically, epoxy-based SMPs are interesting because of their excellent mechanical properties, electrical insulation and high thermal stability, being potentially
Polymers 2015, 7, 2146–2164; doi:10.3390/polym7101505
www.mdpi.com/journal/polymers
Polymers 2015, 7, 2146–2164
interesting in structural applications [3,6]. Nevertheless, the inherent rigidity of common epoxy resins, their low strain and stress at break, narrows their application range. Many researchers employ curing agents characterized by flexible and long aliphatic chains to overcome the deformability limitations. Wei et al. [7] enhanced the segmental mobility by introducing a long flexible chain of poly(propylene glycol) diglycidyl ether (PPGDGE) in epoxy-amine formulations. Rousseau et al. [8] used curing agents of different nature (aliphatic-based, propylene oxide (PO)-based, ethylene oxide (EO)-based, PO/EO) on a DGEBA (epoxy resin) and studied the influence of the network structure on the ultimate strain and the impact on the shape-recovery. The results indicated that by carefully varying the network structure it was possible to increase the tensile elongation while maintaining high values of glass transition temperature (Tg ) and obtaining excellent shape-recovery performances. On the other hand, another approach to enhance the ultimate strain based on the effect of the programming temperature (Tprog ) was investigated by Yakacki et al. [9] working with acrylates and Rousseau et al. [10] using epoxy-based systems. The results demonstrated an improvement of the strain at break of three- to five-fold, when deforming at a temperature slightly below the Tg , corresponding to the onset point of the network relaxation process (measured by dynamic-mechanical analyses). Recently, Zheng et al. [11] presented a high strain two component epoxy-amine (up to 210% of strain at break when deforming at the onset temperature) with tailorable Tg and excellent shape-fixity and shape-recovery performances. Nevertheless, while the ultimate strain is highly enhanced by varying the network structure, both the ultimate strength and the shape-recovery process are affected by these changes. In structural and aerospace applications [12,13] the capability to perform a mechanical work and the control of the shape-recovery process are highly demanded to properly tailor materials for specific conditions. However, one of the major drawbacks is to fulfil both, high ultimate strength and strain. Moreover, the enhancement of the strain limits leads to uncontrolled modifications of the shape-recovery process. To this end, very few works have focused on the study of the relation between the network structure and, either the strength or the shape-recovery process. Williams et al. [14] improved the tensile strain of an epoxy-based system and reported a great strength by combining chemical and physical crosslinking through a long aliphatic chain (n-dodecylamine) and a compact rigid structure (m-xylylenediamine). Our group developed hyperbranched poly(ethyleneimine)-modified epoxy SMPs with enhanced ultimate strength (up to 12 MPa of stress at break), maintaining high ultimate strain [15]. On the other hand, Yakacki et al. [16,17] studied the dependence of the shape-recovery process on the network structure in constrained and unconstrained (stress-free) conditions. The results evidenced the relationship between the network relaxation dynamics and the unconstrained (stress-free) recovery process, and the dependence on the crosslinking density for the constrained ones. Chen et al. [18] evidenced the influence of the network relaxation dynamics in unconstrained recovery processes by means of numerical simulations. However, the enhanced ultimate strength and strain reported are not sufficiently high, and the knowledge of the network structure role on the shape-recovery process is not yet well-defined. More efforts to better understand the relation of the network structure properties with the shape-recovery process and ultimate mechanical properties are required to advance in the development of smart materials. The so-called click chemistry is a new generation of reactions based on efficiency, versatility and selectivity [19–21]. Among the different “click” mechanisms, the thiol-epoxy systems [22] are especially interesting due to the formation of hydroxyl and thioether groups in a single step during the polymerization process, which can be further transformed into other polymeric structures [23,24]. Tertiary amines are commonly used as base catalysts in thiol-epoxy formulations, but they are usually too reactive, making it hard to control the reaction and shortening their pot-life. It was recently shown [25,26] that it is possible to increase the pot life of thiol-epoxy formulations by means of polymer-encapsulated amines or thermally stable aromatic-aliphatic ureas, being the former the most effective way. Thus, it is possible to prepare formulations, store them safely before application,
2147
Polymers 2015, 7, 2146–2164
and process them at temperatures high enough to quickly release the amine catalyst and produce a completely cured material in a short period of time. Over the last few years, some reports have been published on the use of the “click” concept to synthesize SMPs. Shandas et al. [27] used UV-induced thiol-ene polymerization to develop SMPs with a wide range of mechanical and thermo-mechanical properties with excellent SMR. Afterwards, Sunitha et al. [28] and McBride et al. [29] synthesized SMPs based on azyde-alkine–CuAAC “click” chemistry with high shape-recovery and shape-fixation performances. To the best of our knowledge, though, the “click” chemistry via nucleophilic attack to strained heterocycles (i.e., the thiol-epoxy mechanism), using thermal latent initiators, has not been yet investigated on the development of SMPs. In this paper, we combine the advantages of using “click” thiol-epoxy reaction with latent catalysis to produce SMPs with enhanced mechanical properties (high ultimate strength and strain) in a simple and efficient manner. This study expects to provide, by means of a detailed thermal and mechanical characterization of these materials, a better understanding on the relationship between the network structure and thermo-mechanical properties with the SMR, and to establish sound criteria for material choice depending on the expected application or certain operation conditions. In this sense, it is expected to clarify as much as possible the parameters one should consider depending on the desired applications. On the other hand, by varying the programming temperature, it is intended to link both the enhancement on the mechanical properties [9,10] with the effect on the SMR. For this purpose, in this work, (3- and 4- functional) thiol components have been used as curing agents for a commercial diglycidyl ether of bisphenol A (DGEBA) via “click” chemistry using encapsulated amines as latent initiators. In addition, different amounts of triglycidyl isocyanurate (iso) have been used as network structure modifiers in order to increase the Tg and the crosslinking density of the material. The SMR has been analyzed using different programming temperatures (below, at and above the Tg ), and has been related to the network structure changes induced by the addition of isocyanurate and the use of thiol components of different functionality. 2. Experimental Section 2.1. Materials A commercial epoxy resin, Diglycidyl ether of bisphenol A (DGEBA, GY240, Huntsman, Everberg, Belgium) with a molecular weight per epoxy equivalent unit of 182 g/equiv was used as the main epoxy resin. Pentaerythritol tetrakis (3-mercaptopropionate) (4thiol), with a molecular weight per thiol equivalent unit of 122.17 g/equiv (Sigma-Aldrich, St. Louis, MO, USA) and trimethylolpropane tris(3-mercaptopropionate) (3thiol), with a molecular weight per thiol equivalent unit of 132.85 g/equiv (Sigma-Aldrich, St. Louis, MO, USA) were used separately as crosslinking agents. In addition, a tri(2,3-epoxypropyl)isocianurate (iso), with a molecular weight per epoxy equivalent unit of 99.09 g/equiv (Sigma-Aldrich, St. Louis, MO, USA) was used as a modifier in different weight proportions: 10%, 20%, 30% and 40% (i.e., 10:90 wt % ISO:DGEBA), maintaining a stoichiometric proportion of thiol and epoxy groups (epoxy:thiol ratio equal to 1:1). An encapsulated imidazole (LC80, AC Catalysts) was used as a latent initiator in a proportion of 0.5 phr (parts of catalyst for hundred parts of epoxy resin). The reagents were used as received, without further purification. The formulations were prepared by manually stirring the components in a glass vial and carefully pouring the mixtures in an open teflon mold. The curing process was carried out in an oven, one hour at 120 ˝ C followed by one hour at 150 ˝ C to allow the completion of the process. Scheme 1 shows the expected network for the mixtures without iso and Scheme 2 the expected changes introduced by the iso (the crosslinking points have been highlighted).
2148
Polymers 2015, 7, 2146–2164 Polymers 2015, 7, page–page Polymers 2015, 7, page–page O
O
O
O
O
O
O
O OH O
DGEBA DGEBA
+ +
(latent amines LC-80, Δ precursor) (latent amines precursor)
O HS
O O
HS HS
O O
HS
O
LC-80, Δ
O
S
O O
S
O O
O
S
O O
S
Ocrosslinking point O crosslinking point S
OH
O O
SH
O
O
SH
O
OH
O
O S
O
O
S
O
S
O O
O
O
O O
O O 3thiol O
S
O O
S
OH
O
3thiol
S S
O
Scheme network structure structure of ofthe themixture mixturecontaining containing3thiol 3thiolasascuring curing Scheme1.1.Reaction Reaction scheme scheme and expected network agent, as resin and LC-80 initiator (3thiol-DGEBA-LC80) formulation. agent,DGEBA DGEBA asepoxy epoxy resin and LC-80 as as (3thiol-DGEBA-LC80) formulation. Scheme 1. Reaction scheme and expected network structure of the mixture containing 3thiol as curing agent, DGEBA as epoxy resin and LC-80 as initiator (3thiol-DGEBA-LC80) formulation. O
O OH O
OH
S S
O
O O
S
S
O
O
O
S
O
O
O
S
O
crosslinking point O O OH O O
OH
S S
O O O
O
crosslinking point HO O HO O O S O O
O O
S
O
O
N
N
N N
N
O O
O
O O
O N N N O O
O
N
O
N O N
S
O
N
O
iso
O
iso
O
S
O Scheme 2. Expected network structure of the 3thiol-DGEBA-LC80 modified with the (iso) component.
Scheme 2. 2. Expected Scheme Expected network network structure structure of of the the 3thiol-DGEBA-LC80 3thiol-DGEBA-LC80 modified modifiedwith with the the (iso) (iso) component. component.
The composition of the different formulations considered is detailed in Table 1. The composition of the different formulations considered is detailed in Table 1. 1. Composition of the different formulations of study. The composition ofTable the different formulations considered is detailed in Table 1.
Table 1. Composition of the different formulations of study. Formulation (wt %)of theThiol (wt %) ISO (wtof%) Table 1.DGEBA Composition different formulations study.LC80 (wt %) 3thiol-NEAT 57.64 42.07 Formulation DGEBA (wt %) Thiol (wt %) ISO0.00 (wt %) LC800.29 (wt %) 3thiol-(10%)iso 50.11 44.04 5.57 0.28 3thiol-NEAT 57.64 42.07 0.00 0.29 Formulation DGEBA (wt %) Thiol (wt %) ISO (wt %) LC80 (wt %) 3thiol-(20%)iso 43.08 45.88 10.77 0.27 3thiol-(10%)iso 50.11 44.04 5.57 0.28 3thiol-NEAT 57.64 42.07 0.00 0.29 3thiol-(30%)iso 36.49 47.61 15.64 0.26 3thiol-(20%)iso 43.08 45.88 10.77 0.27 3thiol-(10%)iso 50.11 44.04 5.57 0.28 3thiol-(40%)iso 30.31 49.22 20.21 0.25 3thiol-(30%)iso 36.49 47.61 15.64 0.26 3thiol-(20%)iso 43.08 45.88 10.77 0.27 4thiol-NEAT 59.66 40.04 0.00 0.300.26 3thiol-(40%)iso 30.31 49.22 20.21 0.25 3thiol-(30%)iso 36.49 47.61 15.64 4thiol-(10%)iso 51.95 41.99 5.77 0.290.25 4thiol-NEAT 59.66 40.04 0.00 0.30 3thiol-(40%)iso 30.31 49.22 20.21 4thiol-(20%)iso 44.73 43.81 11.18 0.280.30 4thiol-(10%)iso 51.95 41.99 5.77 0.29 4thiol-NEAT 59.66 40.04 0.00 4thiol-(30%)iso 37.95 45.52 16.26 0.270.29 4thiol-(10%)iso 51.95 41.99 5.77 4thiol-(20%)iso 44.73 43.81 11.18 0.28 4thiol-(40%)iso 31.56 47.13 21.04 0.260.28 4thiol-(20%)iso 44.73 43.81 11.18 4thiol-(30%)iso 37.95 45.52 16.26 0.27 4thiol-(30%)iso 37.95 45.52 16.26 4thiol-(40%)iso 31.56 47.13 21.04 0.260.27 4thiol-(40%)iso 31.56 47.13 21.04 0.26 2.2. Network Structure and Thermo-Mechanical Properties
2.2. Network Structure and Thermo-Mechanical Properties The network structure and thermo-mechanical properties of the different materials were studied 2.2. Network Structure and Thermo-Mechanical Properties by thermo dynamic-mechanical analyses using a properties DMA Q800 (TAdifferent instrument) equipped with a The network structure and thermo-mechanical of the materials were studied The network structure and thermo-mechanical properties of the different materials were Single-Cantilever (10 mm) clamp. The samples were Q800 thoroughly polished until a prismatic by thermo dynamic-mechanical analyses using a DMA (TA instrument) equipped with a 3 studied by thermo using athoroughly DMA Q800 (TAanalyzed instrument) equipped rectangular shape (10 (≈dynamic-mechanical 20mm) × 7.5clamp. × 1.8 mm was obtained. samples were Hz with Single-Cantilever The) analyses samples were The polished until ata 1prismatic with a Single-Cantilever mm) The samples were thoroughly polished prismatic oscillation amplitude of 10 μm aclamp. heating rateobtained. of 3 °C/min. 1 shows aanalyzed typicaluntil representation rectangular shape (≈ 20 ×(10 7.5 ×at1.8 mm3) was TheFigure samples were at 1a Hz with 3 g) was of the curves obtained by DMA. The T was determined by the peak of the tanδ curve, both, rectangular shape (« 20 ˆ 7.5 ˆ 1.8 mm obtained. The samples were analyzed at 1 Hz with oscillation amplitude of 10 μm at a heating rate of 3 °C/min. Figure 1 shows a typical representation ˝ thethe modulus atobtained room of temperature g′) and above the glass transition (E r′), were determined directly oscillation amplitude 10 at a (E heating ofdetermined 3 C/min. Figure shows typical of curves byµm DMA. The Tg rate was by the 1peak of athe tanδrepresentation curve, both, from the storage modulus curve (E′). The onset point gE′) was determined as the intersection of the the modulus at room temperature (Eg′) and above the (T glass transition (Er′), were determined directly E′ from the storage modulus curve (E′). The onset point 2149 4 (Tg ) was determined as the intersection of the 4
Polymers 2015, 7, 2146–2164
of the curves obtained by DMA. The Tg was determined by the peak of the tanδ curve, both, the modulus at room temperature (Eg 1 ) and above the glass transition (Er 1 ), were determined directly 1 from the2015, storage modulus curve (E1 ). The onset point (Tg E ) was determined as the intersection of Polymers 7, page–page the lines corresponding to the plateau in the glassy state and the modulus fall. In addition, the width of the corresponding tanδ curve at the half-height (FWHM) andstate the tanδ peak have been lines to the plateau in the glassy and the modulus fall.determined In addition,as theshown widthin of Figure 1. curve at the half-height (FWHM) and the tanδ peak have been determined as shown in Figure 1. the tanδ
Figure Figure1.1.Schematic Schematicrepresentation representationof ofthe therelevant relevantparameters parametersobtained obtainedfrom fromDMA DMAanalysis: analysis:storage storage 1 ) and tanδ curve for a typical thermoset. The parameters of interest have been pointed out. modulus modulus(E (E′) and tanδ curve for a typical thermoset. The parameters of interest have been pointed out.
Accordingtotothe theAffine Affinemodel modelfor forideal idealrubbers, rubbers,the therelaxed relaxedmodulus modulusEE1r′at atlow lowstrain strainrates ratesisis According r relatedtotothe thestrand stranddensity densityofofthe thecrosslinked crosslinkednetwork networkby byEquation Equation(1). (1). related (1) 3 ρ Er1 “ 3 ¨ R ¨ T ¨ ρ ¨ ve (1) where R is the universal gas constant, T is the temperature at which the modulus Er′ is calculated, 1 ρ is the of the material at the temperature T and veatis which the network strand Edensity. Usually, where R density is the universal gas constant, T is the temperature the modulus r is calculated, network strand ve at is the considered to depend thethe crosslink functionality as: ρthe is the density of thedensity material temperature T and on ve is networkdensity strand and density. Usually, the network strand density ve is considered to depend on the crosslink density and functionality as: (2) 2˙ ÿˆ f ve “ nf ¨ (2) 2 where nf is the density of crosslinking junctions with functionality f. However, it has been shown that f ě3
the mobility of the network strands and their deformation is further affected by the functionality of where n f is the density crosslinking with However, it so hasthat been shown that the crosslinking pointsof[30] due to thejunctions fluctuation offunctionality crosslinks in f.the network, modifications the mobility of the network strands and their deformation is further affected by the functionality of to the ideal expression have been proposed such as [31]: the crosslinking points [30] due to the fluctuation of crosslinks in the network, so that modifications 2 to the ideal expression have been proposed such as [31]: (3) 2 ˙ ˆ ÿ f ´2 ve “ none f ¨ has to take into account the composition of (3) For the correct evaluation of this parameter, the 2 f ě3 formulation and that, in the fully cured materials, 3thiol and iso contribute with trifunctional crosslinks, 4thiol with tetrafunctional crosslinks. For theand correct evaluation of this parameter, one has to take into account the composition of Other non-idealities should have an the value of contribute Er′, such as thetrifunctional presence of the formulation and that, in the fully curedinfluence materials,in3thiol and iso with permanent physical entanglements [31] or the non-Gaussian deformation of the network strands [32]. crosslinks, and 4thiol with tetrafunctional crosslinks. Lessernon-idealities and Crawfordshould [33] studied theinfluence dependence of the glass temperature on the Other have an in the value of transition Er 1 , such as the presence of crosslinking density and the crosslinking functionality, and proposed the following modification of permanent physical entanglements [31] or the non-Gaussian deformation of the network strands [32]. the classical free-volume based relationship taking into account the effect of crosslink fluctuation, as: ,
2 2150
2
(4)
where Tg,x is the glass transition temperature at a conversion x, fave is the average crosslink functionality, ζ is proportional to the molecular weight of the unreacted resin and to the ratio of
Polymers 2015, 7, 2146–2164
Lesser and Crawford [33] studied the dependence of the glass transition temperature on the crosslinking density and the crosslinking functionality, and proposed the following modification of the classical free-volume based relationship taking into account the effect of crosslink fluctuation, as: Tg,x “ Tg8 `
2 ¨ p f ave ´ 2q ζ ¨ f ave Mc
(4)
where Tg,x is the glass transition temperature at a conversion x, f ave is the average crosslink functionality, ζ is proportional to the molecular weight of the unreacted resin and to the ratio of incremental free volume contributions from the resin and the curing agent, and Mc is the average Polymers 2015, 7, page–page mass between crosslinks. This expression can be conveniently rearranged and adapted for the present case of comparison between similar materials with differing crosslinking density and functionality. mass between crosslinks. This expression can be conveniently rearranged and adapted for the present It can be shown that the Tg is approximately proportional to the an effective crosslinking density, in case of comparison between similar materials with differing crosslinking density and functionality. a similar way to the Fox-Loshaek equation [34] yielding: It can be shown that the Tg is approximately proportional to the an effective crosslinking density, in a similar way to the Fox-Loshaek equation [34] T “yielding: a`b¨v (5) g
e
(5) where a and b are constants and ve is calculated using expression (3), taking into account is calculated using expression (3), taking into for account where a fluctuation and b are and constants and ve Other crosslink functionality. expressions have been proposed to account other crosslink fluctuation and functionality. Other expressions have been proposed to account for other non-idealities [35,36] but they are not considered for simplicity purposes. non-idealities [35,36] but they are not considered for simplicity purposes. 2.3. Mechanical Characterization 2.3. Mechanical Characterization The mechanical response was analyzed through stress-strain experiments using the DMA Q800 The mechanical response throughwith stress-strain experiments using the DMA Q800 (TA instrument, New Castle, was DE, analyzed USA) equipped a Tension-Film clamp in Force-controlled (TA instrument, Castle, DE, USA) equipped a Tension-Film clamp in Force-controlled mode at mode at a rangeNew of temperatures (from Troom towith Tg + 20). Dog-bone shaped samples were polished athoroughly range of temperatures (from T room to T g + 20). Dog-bone shaped samples were polished thoroughly with sandpaper in order to obtain controlled size samples (length ˆ thickness ˆ width 3 , to with in mm order controlled size samples (length × thickness 15 an × 0.5 × 1.4before mm3, « 15sandpaper ˆ 0.5 ˆ 1.4 seeobtain Figure 2) and were annealed during 10 min at× width 120 ˝ C≈ in oven see FigureThe 2) and were were annealed duringand 10 min at 120held °C inbetween an oven the before testing. samples were testing. samples mounted carefully grips of theThe clamp to ensure mounted and carefully held between the grips of the clamp to ensure an initial testing length an initial testing length of 10–11 mm. The pressure applied in both grips was sufficiently low of to 10–11 mm. The applied in bothhigh grips lowexperiments to avoid cutting the sample, avoid cutting thepressure sample, and sufficiently to was avoidsufficiently slipping. The were carried out at and sufficiently high and to avoid experiments were carried out at 3 N/min. The stresspoint and 3 N/min. The stress strainslipping. at breakThe (σb and εb respectively) were determined as the failure strain at break (σ b and ε b respectively) were determined as the failure point from the curves obtained. from the curves obtained.
Figure 2. Dog-bone shaped samples preparation process. Figure 2. Dog-bone shaped samples preparation process.
2.4. Shape-Memory Response (SMR) 2.4. Shape-Memory Response (SMR) The SMR of all the formulations was investigated through a visual methodology (qualitative study) The SMR of all the formulations was investigated through a visual methodology (qualitative for comparison purposes and as a means of choosing the best candidates for further analyzing study) for comparison purposes and as a means of choosing the best candidates for further analyzing the effect of the programming temperature in the SMR (quantitative study). the effect of the programming temperature in the SMR (quantitative study). 2.4.1. Qualitative Study The SMR was analyzed using a visual methodology to monitor isothermal-recovery processes. Prismatic rectangular samples of 40 × 8 × 2.5 mm3 (length × width × thickness) were polished to flat parallel surfaces. The programming process was performed following the next steps: the sample 2151(Tprog) stipulated as its Tg to make possible the was heated up to the programming temperature comparison among the different formulations under study, and then deformed to a U-shape using the bending device shown in Figure 3a. The device and the sample were rapidly cooled down with
Polymers 2015, 7, 2146–2164
2.4.1. Qualitative Study The SMR was analyzed using a visual methodology to monitor isothermal-recovery processes. Prismatic rectangular samples of 40 ˆ 8 ˆ 2.5 mm3 (length ˆ width ˆ thickness) were polished to flat parallel surfaces. The programming process was performed following the next steps: the sample was heated up to the programming temperature (Tprog ) stipulated as its Tg to make possible the comparison among the different formulations under study, and then deformed to a U-shape using the bending device shown in Figure 3a. The device and the sample were rapidly cooled down with cold water while maintaining the force applied. The sample was unloaded and was placed in an oven at the isothermal recovery temperature (Trec ) corresponding to its Tg (following the same line of reasoning as for the Tprog ) as shown in Figure 3b. The process was recorded using a high resolution camera and further analyzed frame by frame (one frame per second of the video), measuring the angles as shown in Figure 3b and applying Equation (6) on each frame. The time needed to recover from the 15% to the 85% of the original shape (tsr ) was determined using the Equation (7) (the beginning and the end of the process were neglected to avoid possible experimental errors 2015, since7,itpage–page is a visual methodology). Polymers
Figure 3. Schematic representation of the bending device; (a) and the shape-recovery scenario; Figure 3. Schematic representation of the bending device; (a) and the shape-recovery scenario; (b) for the isothermal-recovery experiments. (b) for the isothermal-recovery experiments.
∆∅ % 100 ∅ ∆Φ∅ t 180° p%SRqΦ “ ¨ 100 ˝ 180 ´ Φ0
(6) (6)
(7)(7) tsr “ t% p%SRq85 ´ tp%%SRq15 where where%SR %SRisisthe theshape-recovered shape-recovered in in percentage, percentage, ΦΦ00isisthe theinitial initialangle angle and and ∆Φ ∆Φt tisisthe thedifference difference between the angle at a time t, Φ t, and Φ0. tsr is the shape-recovery time. between the angle at a time t, Φt , and Φ . tsr is the shape-recovery time. 0
2.4.2. 2.4.2.Quantitative QuantitativeStudy Study InInthe thequantitative quantitativestudy, study,the theSMR SMRwas wasanalyzed analyzedthrough throughaacustomized customizedexperimental experimentalmethod method using the DMA Q800 (TA Instrument) equipped with the Tension-Film clamp. The samples using the DMA Q800 (TA Instrument) equipped with the Tension-Film clamp. The sampleswere were prepared preparedininthe thesame sameway wayas asfor fortheir theirmechanical mechanicalcharacterization characterization(see (seeFigure Figure2). 2).The Theexperiments experiments 1 were 4 three Tprog were chosen: TgE′, TTggEand + 20, wereperformed performedfollowing followingthe thesteps stepsshown shownininFigure Figure 4 three Tprog were chosen: , TgTgand Tg 1 E′) [9,10]Eand along the glass transition process to+quantify the SMR at the optimal mechanical point (T g 20, to quantify the SMR at the optimal mechanical point (Tg ) [9,10] and along the glass transition (Tprocess g and Tg(T +g20). were heated were up to heated Tprog and 5 min this temperature to andThe Tg samples + 20). The samples upheld to Tduring heldatduring 5 min at this prog and ensure thermaltoequilibrium (point 1 in Figure(point 4). A 1uniaxial tensile at a constant forceat temperature ensure thermal equilibrium in Figure 4). Adeformation uniaxial tensile deformation rate of 1 N/min applied until theapplied 75% ofuntil εb (εthe D) was reached a to point 2 in a constant force was rate of 1 N/min was 75% of εb (εD )(line was reached (line a toFigure point 24).in This limit of 75% of εD was used in order to make comparisons between the formulations regardless of their different network structure and mechanical 2152properties, also depending on Tprog, as well as to quantify the SMR near the maximum elongation allowed. Once εD was reached, the force applied was held constant while rapidly cooling to setting temperature (Ts) (line b to point 3). Subsequently, the force applied was released at the same force rate of 1 N/min (line c to point 4). Afterwards,
Polymers 2015, 7, 2146–2164
Figure 4). This limit of 75% of εD was used in order to make comparisons between the formulations regardless of their different network structure and mechanical properties, also depending on Tprog , as well as to quantify the SMR near the maximum elongation allowed. Once εD was reached, the force applied was held constant while rapidly cooling to setting temperature (Ts ) (line b to point 3). Subsequently, the force applied was released at the same force rate of 1 N/min (line c to point 4). Afterwards, a heating rate of 3 ˝ C/min was imposed until the original shape was recovered (line d to point 5). All7,the other experimental conditions and parameters remained unchanged. Polymers 2015, page–page
Figure Figure4. 4.Thermo-responsive Thermo-responsiveSMPs; SMPs; 3D 3D representation representation of of the the programing programing process. process.
The SMR was analyzed as the average of the results through three consecutive cycles (it was not The SMR was analyzed as the average of the results through three consecutive cycles (it was not appreciated any worsening of the SMR during the cycling until the sample broke). The capability of appreciated any worsening of the SMR during the cycling until the sample broke). The capability of the material to recover its original shape was determined by the shape-recovery ratio (Rr), using the material to recover its original shape was determined by the shape-recovery ratio (R ), using the the Equation (8) and the capability to retain the imposed shape (temporary shape) was rdetermined Equation (8) and the capability to retain the imposed shape (temporary shape) was determined by by the fixation-ratio (Rf), using Equation (9). the fixation-ratio (R f ), using Equation (9). ε ε 100 R % (8) εDε´ εp Rr p%q “ ¨ 100 (8) ε εD R % 100 (9) εεu Rf p%q “ ¨ 100 (9) ε where εD is the deformation reached after loading, Dεu is the deformation after unloading and εp is where εD is thestrain deformation after loading, εu is the deformation after unloading and εp4.is the the permanent after thereached shape-recovery. All these parameters are represented in Figure permanent strain after the shape-recovery. All these parameters are represented in Figure 4. The shape-recovery rate (Vr) was calculated using the Equations (10) and (11), enclosing the shape-recovery rate (V r )ofwas usingrecovered) the Equations (10) and the (11),early enclosing the wholeThe process (from 1% to 100% the calculated original shape and avoiding and final whole (from process (from 1% to 100% of the original shape recovered) and avoiding the early and final stages 15% to 85%). stages (from 15% to 85%). ε ε ε ´ εp 100 %SR 100 (10) %SR “ 100 ´ ε (10) ε ¨ 100 εu ´ εp ˝ VVr p%{ Cq “ %/°C
∆%SR ∆%SR ∆T∆% ∆T ∆%SR
(11) (11)
where %SR %SR is is the the percentage percentage of of original original shape shape recovered recovered normalized normalized to to the the permanent permanent strain strain (ε (εpp),), where ∆%SR is the range tested (%SR = 1%–100% and 15%–85%) and ∆T is the temperature increment ∆%SR Δ%SR is the range tested (%SR = 1%–100% and 15%–85%) and ΔTΔ%SR is the temperature increment corresponding to to the the Δ%SR. ∆%SR. corresponding
3. Results and Discussion 2153 3.1. Thermo-Mechanical and Network Structure Properties
The thermo-mechanical and network structure properties were analyzed by dynamic-mechanical analyses. The results for the 3thiol and 4thiol systems are shown in Figures 5 and 6, and some
Polymers 2015, 7, 2146–2164
3. Results and Discussion 3.1. Thermo-Mechanical and Network Structure Properties The thermo-mechanical and network structure properties were analyzed by dynamic-mechanical analyses. The results for the 3thiol and 4thiol systems are shown in Figures 5 and 6 and some characteristic parameters of the network relaxation process (see Figure 1) are Polymers 2015, 7, in page–page summarized Table 2. Polymers 2015, 7, page–page
Figure 5. Dynamic-mechanical analyses for the (3thiol) different formulations. Figure 5. Dynamic-mechanical analyses for the (3thiol) different formulations. Figure 5. Dynamic-mechanical analyses for the (3thiol) different formulations.
Figure 6. Dynamic-mechanical analyses for the (4thiol) different formulations. Figure 6. Dynamic-mechanical analyses for the (4thiol) different formulations.
6. Dynamic-mechanical analyses forlower the (4thiol) different formulations. In general,Figure the formulations with 3thiol have Tg than those with 4thiol because of the reduction of free volume and hindrance of the molecular motion caused the presence In general, the formulations with 3thiol have lower Tg than those withby4thiol because of of In general, the formulations with 3thiol have lower T than those with 4thiol because of[33], the g crosslink junctions [36],volume which and becomes more relevant with increasing functionality the reduction of free hindrance of the molecular motioncrosslink caused by the presence of reduction ofother free volume and hindrance of [37,38]. the molecular motion causedadding by thethe presence of crosslink as seen for shape-memory materials On the other hand, iso compound to crosslink junctions [36], which becomes more relevant with increasing crosslink functionality [33], junctions [36], which tend becomes more relevant with increasing functionality [33], as seen the different systems to increase the T[37,38]. g. According the crosslink expected network Scheme 2, as seen for other shape-memory materials On thetoother hand, adding the of isothe compound to for other shape-memory materials [37,38]. On the other hand, adding the iso compound to the the iso component acts as a 3-functional crosslinking point, increasing the functionality of the system, the different systems tend to increase the Tg. According to the expected network of the Scheme 2, different systems tend to increase the TgNevertheless, . According tothe theTexpected network of become the Scheme 2, the iso thus, creating a more hindered network. g increment tends to less the iso component acts as a 3-functional crosslinking point, increasing the functionality of therelevant system, component acts as 3-functional point, increasing the functionality of the system, thus, especially above anaiso content network. upcrosslinking to 30%Nevertheless, for formulations 3thiol. Figure 7 shows a good thus, creating a more hindered the Tcontaining g increment tends to become less relevant correlation between g and the effective strand density ve (calculated taking into account crosslink especially above anthe iso Tcontent up to 30% for formulations containing 3thiol. Figure 7 shows a good functionality and fluctuation, Equation (3)), showing a fairly linear trend as predicted by 2154 correlation between the Tg and using the effective strand density ve (calculated taking into account crosslink Equation (5). However, some non-idealities appear for both a3thiol 4thiol formulations with functionality and fluctuation, using Equation (3)), showing fairlyand linear trend as predicted by increasing iso content and, consequently, v e. It must be taken into account that the increase in the Equation (5). However, some non-idealities appear for both 3thiol and 4thiol formulations with content of iso also changes the type of network structure, introducing crosslinks with a non-negligible
Polymers 2015, 7, 2146–2164
creating a more hindered network. Nevertheless, the Tg increment tends to become less relevant especially above an iso content up to 30% for formulations containing 3thiol. Figure 7 shows a good correlation between the Tg and the effective strand density ve (calculated taking into account crosslink functionality and fluctuation, using Equation (3)), showing a fairly linear trend as predicted by Equation (5). However, some non-idealities appear for both 3thiol and 4thiol formulations with increasing iso content and, consequently, ve . It must be taken into account that the increase in the content of iso also changes the type of network structure, introducing crosslinks with a non-negligible Polymers 2015, 7, and page–page mass, the size and rigidity of the network strands [25].
1
7. (Tg ) and (Er ) with respect to the effective strand density (ve ) corresponding to each % of Figure 7. (TgFigure ) and (E r′) with respect to the effective strand density (ve) corresponding to each % of iso content. iso content.
FWHM and tanδ peak values are characteristics of the relaxation and can be related to the FWHM and tanδ peak values are characteristics of the relaxation and can be related to the crosslinking density and and the the network the material. is commonly observed that crosslinking density networkhomogeneity homogeneity ofofthe material. It isItcommonly observed that the relaxation peakspeaks broaden andand decrease with increasing crosslinking [15] the relaxation broaden decrease its its intensity intensity with increasing crosslinking density density [15] a phenomenon is inherently related the increase in crosslinking density andTgT[34]. a phenomenon that is that inherently related to thetoincrease in crosslinking density and Although g [34]. Although the trend is not well defined, because of the discrete increase in crosslinking density, the the trend is not well defined, because of the discrete increase in crosslinking density, the same is same is observed in the present case. The effect of iso on the shape of the relaxation in 4thiol systems observed in the present case. The effect of iso on the shape of the relaxation in 4thiol systems is very is very limited in comparison with 3thiol and, in fact, the broadening of the relaxation when 40% of limited in withis 3thiol and, fact, broadening ofonthe when 40% of iso is isocomparison is used with 3thiol too large and in may havethe a detrimental effect therelaxation SMR of the material. used with 3thiol is too large and may have a detrimental effect on the SMR of the material. Table 2. Network structure and thermo-mechanical properties of the different materials of study. 2155
Formulation
Eg′ (MPa)
Er′ (MPa)
Eg′/Er′
FWHM (°C)
tanδ peak
Tg (°C)
3thiol-NEAT 3thiol-(10%)iso
1,590 1,804
6.7 8.5
237 211
10.3 11.9
1.28 1.34
44.1 49.1
TgE′ (°C) 37.7 42.0
ve (mol/kg) a 0.530 0.649
Polymers 2015, 7, 2146–2164
Table 2. Network structure and thermo-mechanical properties of the different materials of study. Formulation 3thiol-NEAT 3thiol-(10%)iso 3thiol-(20%)iso 3thiol-(30%)iso 3thiol-(40%)iso 4thiol-NEAT 4thiol-(10%)iso 4thiol-(20%)iso 4thiol-(30%)iso 4thiol-(40%)iso
1
Eg 1 (MPa)
Er 1 (MPa)
Eg 1 /Er 1
FWHM (˝ C)
tanδ peak
T g (˝ C)
T g E (˝ C)
ve (mol/kg) a
1,590 1,804 1,540 1,616 1,744 1,850 1,815 1,855 1,870 1,800
6.7 8.5 9.2 8.5 6.4 10.8 12.6 15.5 17.0 18.2
237 211 167 191 271 171 145 120 110 99
10.3 11.9 11.7 12.5 15.7 12.0 12.1 12.1 11.8 12.6
1.28 1.34 1.31 1.18 1.13 1.06 1.11 1.03 1.05 0.98
44.1 49.1 53.1 56.2 57.9 59.7 66.1 70.6 75.4 77.4
37.7 42.0 46.3 46.6 47.9 52.0 58.8 62.9 68.2 69.7
0.530 0.649 0.760 0.863 0.961 0.822 0.960 1.089 1.209 1.323
a
Calculated using Equation (3).
The effective strand density ve (calculated taking into account crosslink functionality and fluctuation, using Equation (3)) shows an increasing trend with the addition of iso and with the increase in the thiol functionality from 3thiol to 4thiol, as expected. The measured values of the relaxed modulus Er 1 have a very good agreement with this trend except 3thiol formulations with 30% and 40% of iso, that show decreasing values of Er 1 . Such discrepancies are in line with the non-idealities observed in the Tg and the broadening of the relaxation peak in 3thiol formulations with higher iso content. Differences between 3thiol and 4thiol formulations may be consequence of the different structure of both crosslinking agents. Setting aside the different functionality, 3thiol has a pendant ethyl group that may produce a certain distortion in the network structure and the network relaxation dynamics, thus affecting the whole of the measured parameters. In addition, one must consider that the introduction of iso produces not only an increase in crosslinking density but also a significant change in the network structure. Such considerations fall out the scope of the simple theoretical expressions used in this work. Furthermore, it is possible that this also results from a certain incompatibility between iso and 3thiol or a difference in the reactivity of the epoxy groups of iso and DGEBA, leading to an inhomogeneous distribution of the different types of network strands and crosslinking points. 3.2. SMR Qualitative Study: Comparison with Thermo-Mechanical Properties and Network Structure From the previous thermo-mechanical analysis, a series of general considerations can be made in terms of the SMR of the different materials. The capability of a material to show shape-recovery is usually quantified by the Eg 1 /Er 1 ratio [15]. Table 2 shows that the Eg 1 /Er 1 ratio is over 2 orders of magnitude or above for all the formulations, a high enough value. The drop of the modulus from Eg 1 to Er 1 during the network relaxation is steep enough to expect high rates during the shape-recovery process (see Figures 5 and 6). In addition, there is a plateau zone of Eg 1 at the Ts (temperature at which the temporary shape is fixed) sufficiently away from the modulus drop zone to avoid premature shape-recovery and consequently allow good shape-fixation. Overall, the shape of the network relaxation process, represented by the tanδ curve, is well-defined along the glass transition (no noise is noticed) and it is similar for all the formulations, expecting a similar SMR. In the qualitative study, the SMR has been analyzed through the tsr parameter (see Equation (7)), since no losing of the temporary shape was visually appreciated after cooling and releasing the stress applied to the sample. In addition, the programmed samples showed good retention of the temporary shape during the 5–10 min of holding time at room temperature before the shape-recovery was triggered inside the oven. Moreover, at the end of the process, the recovery of the original shape was quantitative %SR « 100% (Equation (8)), leaving no appreciable permanent strain. As it can be seen in Figure 8, overall the tsr is excellent for all the formulations as expected from the Eg 1 /Er 1 ratio and the sharp relaxation. The materials are capable of recovering from the 15% to the 85% of their original shape in less than 14 s when heated on air (note that, in many applications such as security devices, the heating process is on air through isothermal conditions). The NEAT
2156
Polymers 2015, 7, 2146–2164
formulations show a similar response, about 12 s of tsr , while the addition of 10% of iso accelerates the shape-recovery process for both systems (tsr falls to 10–9 s). In contrast, increasing the amount of iso from 10% to 40% slows down the process in the case of the 3thiol formulations (tsr increases from 10 s at 10% of iso up to 13 s at 40% of iso), while for the 4thiol ones it remains almost constant. Figure 8 compares the tsr along with the tanδ peak (a) and FWHM (b) values in order to relate the SMR with the network structure. As it can be seen, tsr decreases when the tanδ peak increases and when FWHM decreases. These results point at a strong relationship between the sharpness of the relaxation process and the rate of the shape-recovery process. Materials with a more heterogeneous network structure and a broader relaxation profile (higher FWHM and lower tanδ peak) have a slower relaxation process, thus limiting the chain motion and shape-recovery. This trend is well-appreciated in the case of 3thiol formulations between 20% and 40% of iso, showing an increase in FWHM from 11.7 to 15.7 ˝ C and an increase in tsr from 11 to 13 s. In the case of the 4thiol formulations this trend is also observed from 30% to 40% of iso. The value of the tanδ peak clarifies the situation when FHWM values are similar. In Figure 8c it is seen that tsr decreases when the tanδ peak increases showing a linear correlation for both the 3thiol and 4thiol systems with isocyanurate. In contrast, other network structure parameters, such as the crosslinking density given by ve or Er 1 , do not have an influence the recovery rate. As discussed in the preceding section, ve increases with the amount Polymerson 2015, 7, page–page of iso (see Table 2), but changes in tsr follow the trend dictated by FWHM and tanδ peak instead, the amount of iso (see Table 2), but changes in tsr follow the trend dictated by FWHM and tanδ peak regardless of the values of ve or Er 1 . These results are in agreement with Yackaki et al. suggesting instead, regardless of the values of ve or Er′. These results are in agreement with Yackaki et al. that on suggesting unconstrained experiments the main responsible the shape-recovery process is the network that on unconstrained experiments the main of responsible of the shape-recovery process relaxation process [17]. is the network relaxation process [17].
Figure 8. Relation between the shape-memory response (in terms of tsr) and the network relaxation Figure 8. Relation between the shape-memory response (in terms of tsr ) and the network relaxation dynamics; (a) tsr against tanδ peak as a function of the composition; (b) tsr against FWHM as a function dynamics; (a) tsr against tanδ peak as a function of the composition; (b) tsr against FWHM as a of the composition; (c) tanδ peak against the tsr for both, 3thiol and 4thiol with isocyanurate functionformulations of the composition; (c) and tanδ40%). peak against the tsr for both, 3thiol and 4thiol with isocyanurate (10%,20%,30% formulations (10%,20%,30% and 40%).
To sum up, these results evidence the dependence of the shape-recovery with the network relaxation on unconstrained experiments and suggesting that FWHM and tanδ peak can be used to To sum up, these results evidence the dependence of the shape-recovery with the network estimate, at least from a comparative point of view, the velocity of the shape-recovery process [39]. relaxation onthe unconstrained experiments and suggesting and tanδ peak can be used to From shape of the modulus curve (the steepness of the that dropFWHM and the plateau zone at the glassy state), one can also obtain a qualitative view of the speed of the recovery process and the capability of the material to retain the temporary shape. 2157 by applying the next criteria: the higher glass Finally, the best candidates have been chosen transition reached by adding the iso component to maximize temporary shape retention along with a homogeneous network structure to have fast recovery process. Therefore, the 3thiol-30%iso and 4thiol-30%iso formulations, along with the NEAT ones, have been chosen for further analysis in the
Polymers 2015, 7, 2146–2164
estimate, at least from a comparative point of view, the velocity of the shape-recovery process [39]. From the shape of the modulus curve (the steepness of the drop and the plateau zone at the glassy state), one can also obtain a qualitative view of the speed of the recovery process and the capability of the material to retain the temporary shape. Finally, the best candidates have been chosen by applying the next criteria: the higher glass transition reached by adding the iso component to maximize temporary shape retention along with a homogeneous network structure to have fast recovery process. Therefore, the 3thiol-30%iso and 4thiol-30%iso formulations, along with the NEAT ones, have been chosen for further analysis in the quantitative study. While there is no doubt in the case of 4thiol formulations, the decision was more complicated in the case of the 3thiol ones. The formulations with 10% and 20% of iso are better candidates from the point of view of the recovery speed, but the formulations with 30% of iso has shown a higher Tg value and its tsr is similar to the NEAT formulation which, overall, is excellent. In addition, as commented above, the Rr and R f , as well as the shape-fixation were excellent for all the formulations. Therefore, for a proper comparative study, the same amount of isocyanurate should be used for both systems. 3.3. Mechanical Influence of the Network Structure and Programming Temperature Polymers 2015,Analysis: 7, page–page One of the most important capabilities of a SMP during the programming process is the 3.3. Mechanical Analysis: Influence of the Network Structure and Programming Temperature mechanical response (in terms of ultimate strain and stress), described by the σb and εb parameters. One of the most capabilities of for a SMP during the programming is the with Figure 9 summarizes the σbimportant and εb values obtained the different formulations andprocess the relation mechanical response (in terms of ultimate strain and stress), described by the σb and εb parameters. the crosslinking density as a function of the Tprog . In general, both parameters have shown a high Figure 9 summarizes the σb and εb values obtained for the different formulations and the relation dependence on temperature and network structure. In accordance with Rousseau and Yackaki [9,10], with the crosslinking density as a function of the Tprog. In general, both parameters have shown a high 1 the optimal mechanical point for the formulations is found atwith theRousseau onset point (Tg E ), [9,10], where the dependence on temperature andall network structure. In accordance and Yackaki material combines both properties: high resistance and molecular mobility, therefore leading E′ the optimal mechanical point for all the formulations is found at the onset point (Tg ), where the to 1 higher values combines of σb andboth εb . As the Tproghigh increases fromand Tg Emolecular to Tg + 20, the σb therefore and εb tend to decrease material properties: resistance mobility, leading to higher values of σb and εb. As weaker the Tprog (i.e., increases from TgE′ to Tg + 20, the σb and εb tend to decrease because of the material becomes for the 4thiol-NEAT system, σb and εb are 39 MPa and E 1 while because of thewhen material becomes weaker the 4thiol-NEAT system, and εb are σ39 MPa εand 66% respectively, programming at T(i.e., programming at Tgσbdecreases g ,for b and b down 66% respectively, when programming at T gE′, while programming at Tg decreases σb and εb down to to 9 MPa and 42%, respectively). On the other hand, the effect of the crosslinking density of the 9 MPa respectively). On appreciated the other hand, the effect crosslinking density the materials materials onand the42%, σb and εb is clearly when Tprogofisthe well above the Tg . Atofthis temperature, on the σb and εb is clearly appreciated when Tprog is well above the Tg. At this temperature, conformational rearrangement of network chains take place quickly in response to the stress applied, conformational rearrangement of network chains take place quickly in response to the stress applied, and therefore the deformation achieved εb is mainly limited by the crosslinking density. As can be and therefore the deformation achieved εb is mainly limited by the crosslinking density. As can be seen, seen, the higher thethe molecular crosslinking points (lower ve ),higher the higher theatstrain the higher molecularweight weightbetween between crosslinking points (lower ve), the the strain at break. trendisisnot not well-defined for σthe σb .explanation The explanation be connected break.InIncontrast, contrast, the the trend well-defined for the b. The must be must connected with with the heterogeneityofofthethe crosslinking points the differences the physical the heterogeneity crosslinking points and and thus thus the differences on the on physical nature ofnature the of chainscreated created among among the [8]. [8]. the chains thedifferent differentformulations formulations
Figure 9. Mechanical response and crosslinking density as a function of the programming
Figure 9. Mechanical response and crosslinking density as a function of the programming temperature for all the formulations of study; (a) strain at break values; (b) stress at break values. temperature for all the formulations of study; (a) strain at break values; (b) stress at break values.
Overall, it should be pointed out the mechanical performance of these materials, in terms of σb and εb, they are superior to other epoxy-based SMPs. Williams et al. [14] reported an epoxy-amine 2158 system (based on a long aliphatic chain n-dodecylamine) with large tensile elongations (up to 75%) and high recovered stresses (over 4 MPa), when programming at a temperature close to the Tg. Rousseau et al. [10] reported epoxy-amine systems (based on long flexible polyetheramines) with E′
Polymers 2015, 7, 2146–2164
Overall, it should be pointed out the mechanical performance of these materials, in terms of σb and εb , they are superior to other epoxy-based SMPs. Williams et al. [14] reported an epoxy-amine system (based on a long aliphatic chain n-dodecylamine) with large tensile elongations (up to 75%) and high recovered stresses (over 4 MPa), when programming at a temperature close to the Tg . Rousseau et al. [10] reported epoxy-amine systems (based on long flexible polyetheramines) with 1 higher values of σb (up to 8 MPa) and large ultimate strain (over 60%), when programming at Tg E , the optimal mechanical point. In addition, our group [15] reported epoxy-amine materials (based on polyetheramines and hyperbranched poly(ethyleneimine)s) with σb over 12 MPa while maintaining large ultimate strain (εb up to 60%). In general terms, while the tensile elongation is enhanced by adjusting the programming temperature and the network structure, the mechanical work remains limited by the low tensile strength obtained. In the present case, the thiol-epoxy materials under study fulfil both the requirements of high tensile strain and enhanced tensile stress, thus, enhancing the energy storage capacity and therefore the expected performance on constrained experiments [40]. The 3thiol-NEAT system reaches values of almost 100% of εb (comparable with those reports by the mentioned authors [10,14,15]) and 55 MPa of σb (which is over 4 times higher than those reported 1 by [15]) when programming at Tg E . Furthermore, when programming at Tg , the σb (over 23 MPa) Polymers 2015, 7, page–page is still higher than those reported by the mentioned authors [10,14,15] and εb is considerably high 1 by [15]) when programming at TgE′. Furthermore, when programming at T σb (over is (up to 82%). The excellent mechanical properties found between Tgg, Ethe and Tg23+MPa) 20 broaden the still higher than those reported by the mentioned authors [10,14,15] and εb is considerably high applicability of our materials, since Tprog not only modifies the mechanical response, but also affects (up to 82%). The excellent mechanical properties found between TgE′ and Tg + 20 broaden the the SMR applicability [41]. In this sense, it is possible to obtain materials with specific and of our materials, since Tprog not only modifies the mechanical response, SMR but alsomaintaining affects excellent mechanical the SMR [41]. response. In this sense, it is possible to obtain materials with specific SMR maintaining and excellent mechanical response.
3.4. SMR Quantitative Study: Influence of the Programming Temperature 3.4. SMR Quantitative Study: Influence of the Programming Temperature
In the quantitative study, the SMR has been investigated through the R and R ratios In the quantitative study, the SMR has been investigated through the Rr and Rrf ratios f (see Equations (8) and(8) (9)). results are Figure (see Equations andThe (9)). The results arerepresented represented inin Figure 10. 10.
10. Recovery-ratio and fixation-ratio function of temperature for all for all the Figure 10.Figure Recovery-ratio and fixation-ratio asasa afunction of the theprogramming programming temperature the formulations of study. formulations of study.
The capability to retain the temporary shape, quantified by the fixation-ratio, Rf, is excellent for the differentto materials with values of almost 100% regardless of the Tprog applied. Rf The capability retain of thestudy temporary shape, quantified by the fixation-ratio, R f The , is excellent for basically depends on the mechanical response of the material during the unloading process, which the different materials of study with values of almost 100% regardless of the T in theapplied. The R f occurs at Ts (the temperature at which the temporary shape is fixed). At Ts the material isprog glassy basically depends onmodulus the mechanical response thefor material during the(see unloading which state and the (Eg′) is higher than 1500ofMPa all the formulations in Table 2)process, thus, is strongly hinderedthe andtemporary the differencesshape in the network structure not affect the occurs at the Ts chain (the mobility temperature at which is fixed). At Tdo the material is in the s mechanical response. Accordingly, the T prog, which probably introduces changes in the network 1 glassy state and the modulus (Eg ) is higher than 1500 MPa for all the formulations (see in Table 2) structure after loading and fixing the temporary shape, does not affect the Rf for the same reason. thus, the chain mobility is strongly hindered and the differences in the network structure do not affect In contrast, the recovery of the original shape, quantified by the recovery-ratio, Rr, depends on the the mechanical response. Accordingly, Tprog , which probably introduces changes the network Tprog: when programming at Tg or Tg +the 20 the values are higher (over 90%–96% at Tg and almost in 100% Tg + loading 20) than those programming at TgE′ (falling to 83% 3thiol-30%iso and the 87% same for structure at after andwhen fixing the temporary shape,down does not for affect the R f for reason. 4thiol-30%iso). Furthermore, the differences in Rr between the different formulations when they are programmed at TgE′ are higher than when they are programmed at Tg + 20. According to Pandini et al. [41], loading at a temperature below the Tg involves more energetic changes (elastic and hardening 2159 processes) while loading at a temperature above the Tg involves mostly entropic changes. The energy lost during the loading process is higher when the loading involves more energetic molecular changes (at TgE′) [42] because of molecular friction. The low initial chain mobility and the activation of the chain relaxation during programming at this temperature, result in some inter-molecular flow
Polymers 2015, 7, 2146–2164
In contrast, the recovery of the original shape, quantified by the recovery-ratio, Rr , depends on the Tprog : when programming at Tg or Tg + 20 the values are higher (over 90%–96% at Tg and almost 1 100% at Tg + 20) than those when programming at Tg E (falling down to 83% for 3thiol-30%iso and 87% for 4thiol-30%iso). Furthermore, the differences in Rr between the different formulations when 1 they are programmed at Tg E are higher than when they are programmed at Tg + 20. According to Pandini et al. [41], loading at a temperature below the Tg involves more energetic changes (elastic and hardening processes) while loading at a temperature above the Tg involves mostly entropic changes. The energy lost during the loading process is higher when the loading involves more 1 energetic molecular changes (at Tg E ) [42] because of molecular friction. The low initial chain mobility and the activation of the chain relaxation during programming at this temperature, result in some inter-molecular flow friction. Part of the deformation work corresponds to viscous flow and cannot be recovered, therefore the energy available for the recovery process is not enough to lead it to completion, resulting in a decrease in Rr . In addition, depending on the nature of the chains 1 (molecular structure), the energy lost during the loading process at Tg E is different. In the case of the NEAT formulations, the Rr is higher and similar for both systems, 3thiol and 4thiol, while adding the isocyanurate decreases the Rr in both systems probably due to the increase in the rigidity of the chains. Moreover, the 3thiol-30%iso shows a lower value of Rr than the 4thiol-30%iso. This must be 2015,previous 7, page–page results of the thermo-mechanical characterization of the materials. Both connected Polymers with the the presence of the side ethyl chain in the structure of 3thiol and some reactivity and compatibility higher and similar for both systems, 3thiol and 4thiol, while adding the isocyanurate decreases the issues, resulting a distortion of tothe network structure andchains. mobility andthea3thiol-30%iso broadening of the Rr in bothin systems probably due the increase in the rigidity of the Moreover, shows a lower value of R r than the 4thiol-30%iso. This must be connected with the previous relaxation, may explain the higher energy losses in the 3thiol formulation with 30% iso.results These results of the characterization the materials. Both the both, presence of mechanical the side ethyl chain suggests that bythermo-mechanical properly varying the Tprog aofbalance enhancing the response and in the structure of 3thiol and some reactivity and compatibility issues, resulting in a distortion of the the shape-recovery capability is possible. network structure and mobility and a broadening of the relaxation, may explain the higher energy In Figure the shape-recovery is analyzed by means of V r (through consecutive losses11, in the 3thiol formulation withprocess 30% iso. These results suggests that by properly varying the3Tprog a balancethe enhancing the mechanical response and of thethe shape-recovery capability is possible. or avoiding cycles), enclosing wholeboth, process (from 1% to 100% original shape recovered), In Figure 11, the shape-recovery process analyzed as by it means Vr (through 3 consecutive the early and final stages (from 15% to 85%). Inisgeneral, wasofexpected from the results of the cycles), enclosing the whole process (from 1% to 100% of the original shape recovered), or avoiding qualitativethe study, all the formulations have shown very high values. If one focuses on the middle early and final stages (from 15% to 85%). In general, as it was expected from the results of the ˝ to 8%/˝ C which are comparable with those reported stage (15%–85%), the values range from 6.3%/ qualitative study, all the formulations have C shown very high values. If one focuses on the middle stage (15%–85%), the values range from 6.3%/°C 8%/°C which are comparable with those reported by [15], thus, allowing almost fully recover of thetooriginal shape in very few minutes. On the other by [15], thus, allowing almost fully recover of the original shape in very few minutes. the other hand, using the average (1%–100%), the effect of both the early and final stages isOn incorporated. As it hand, using the average (1%–100%), the effect of both the early and final stages is incorporated. As it can be seen, the time needed to fully recover the original shape is 2 or 3 times higher. These results can be seen, the time needed to fully recover the original shape is 2 or 3 times higher. These results evidence that the shape-recovery process is islimited theearly early and therefore the evidence that the shape-recovery process limited by by the andand finalfinal stagesstages and therefore the controlstages of theseisstages is crucial to adjust shape-recovery process as desired. control of these crucial to adjust thethe shape-recovery process as desired.
11. Shape-recovery rate ˝(%/°C), enclosing the the whole shape-recovery process (1%–100%) and Figure 11. Figure Shape-recovery rate (%/ C), enclosing whole shape-recovery process (1%–100%) and avoiding the early and final (15%–85%),asasaa function function ofof thethe programming temperature for all avoiding the early and final stagesstages (15%–85%), programming temperature for all the formulations of study. the formulations of study.
On analyzing the effect of the programming temperature with the Vr, it is seen an increment of the Vr with the temperature. This phenomenon is well appreciated if one encloses the whole process (1%–100%) (i.e., the 4thiol-30%iso increases the2160 Vr from 2.5%/°C up to 3.2%/°C and the 3thiol-30%iso from 2.6%/°C up to 3.0%/°C)., This is probably connected with the lower rate of recovery Rr when programming at TgE (see Figure 10). Following the same line of reasoning as above, the lower amount of energy that is effectively stored and available for the recovery process (due to molecular friction losses) decreases both the recovery ratio Rr and the Vr, thus increasing the time needed to fully recover
Polymers 2015, 7, 2146–2164
On analyzing the effect of the programming temperature with the V r , it is seen an increment of the V r with the temperature. This phenomenon is well appreciated if one encloses the whole process (1%–100%) (i.e., the 4thiol-30%iso increases the V r from 2.5%/˝ C up to 3.2%/˝ C and the 3thiol-30%iso from 2.6%/˝ C up to 3.0%/˝ C)., This is probably connected with the lower rate of recovery Rr when programming at Tg E (see Figure 10). Following the same line of reasoning as above, the lower amount of energy that is effectively stored and available for the recovery process (due to molecular friction losses) decreases both the recovery ratio Rr and the V r , thus increasing the time needed to fully recover the original shape. It may be speculated that the resulting stretched network structure of the temporary shape is somewhat altered when programming at Tg E and this may also have an effect on the recovery rate. Whatever the specific cause, this is of particular importance in the case of the 3thiol-30%iso, being the effect of the programming temperature very strong in the central part of the process (15%–85%), and this may be connected with the lowering and broadening of the relaxation described above for the 3thiol-30% (see Table 2 and Figure 8). In spite of the differences in the relaxation process of the different formulations described in the previous sections, the average recovery speed is similar among the different formulations, especially if one encloses the whole process (1% to 100%). The overall trends with respect to the network structure and the shape of the relaxation process are not as clearly defined as in the qualitative study and depend also on the programming temperature. The recovery process in the qualitative study was analyzed under isothermal conditions, using a fixed programming and recovery temperature relative to the network relaxation determined by DMA. Therefore, the time/velocity effect, which is connected with the relaxation dynamics, could be analyzed independently. In contrast, the shape-recovery takes place with increasing temperature in the quantitative study. In this case, there is a time-temperature coupling that adds complexity to the process and makes it more difficult to interpret the average recovery speed data, especially taking into account that there are small differences in terms of relaxation shape but clearly different relaxation temperatures. However, it has been observed that during the recovery process there is indeed a strong effect of the network relaxation dynamics, especially when temperature is still below Tg . There is an excellent correspondence between the early stage of the recovery process and the early stage of the relaxation process determined from the DMA experiments (data not shown). Afterwards, at some point, the temperature enhances the molecular mobility and the recovery process is able to proceed more freely causing acceleration and minimizing the effect of the network structure. Therefore, enclosing the whole process, the effect of the relaxation process is clearly minimized, while focusing on the central part (15%–85%) this effect on the early stage has a major role. To this end, by properly tailoring the network structure and programming temperature, the effect at the early stage of the shape-recovery process can be tuned as desired, obtaining higher or lower rates which at the end of the process will be adjusted. In conclusion, these results evidence the possibility to tailor the early stage and eventually the whole shape-recovery process by properly modifying the network structure and its relaxation dynamics, and the programming temperature. Moreover, the effect on the mechanical properties (stress and strain at break) is also investigated, thus, making possible a balance between mechanical performance and shape-memory performance. 4. Conclusions Materials based on epoxy-thiol “click” chemistry have been obtained, with enhanced mechanical properties and well-defined network structures. The different materials have been characterized by dynamic-mechanical analysis and been tested as SMPs from a qualitative and quantitative point of view. The effect of the material network structure and the programming temperature Tprog on the SMR in unconstrained shape-recovery scenarios has been discussed and analyzed in detail. It has been seen that the network structure of epoxy-thiol materials can be finely tuned by changing both the thiol crosslinker functionality and by introducing a rigid trifunctional epoxy resin.
2161
Polymers 2015, 7, 2146–2164
In general, a good agreement between the experimental data and the expected network structure has been obtained. From the results of the dynamic-mechanical analysis and a qualitative visual methodology for the analysis of the SMR it has been found that the shape-recovery process is highly dependent on the network relaxation dynamics. Thus, the sharpness in the relaxation peak tanδ or the shape of the drop in elastic modulus, can be used to predict, at least from a comparative point of view, the shape-memory response of the materials. Prior to the quantitative analysis, the stress and strain at break values have been determined for a range of temperatures and an optimal point has been found at the onset point of the relaxation process. Values of stress and strain that break as high as 100% and 55 MPa, respectively, have been reached. A strong dependence of the crosslinking density with the strain at break at Tg + 20 has been found, when the network structure rapidly reaches a relaxed state while the deformation is occurring. As the crosslinking density increases, the strain at break decreases. From the quantitative analysis, it has been seen that some parameters such as the ratio of fixation are not affected by programming conditions or network structure, reaching values around 100%. In contrast, the recovery ratio has a strong dependence on both the programming temperature and the network structure: the energy lost during the loading process is significantly reduced due to the decrease in chain friction produced by the network relaxation as the programming temperature is 1 increased. At lower temperatures (Tg E ), the amount of energy lost during the loading process is strongly related to the network structure. Following the same reasoning as above, the shape-recovery rate of the shape-recovery process increases by the reduction of energy lost during the loading process on increasing programming temperature. There is a complex time-temperature superposition during the recovery process with increasing temperature, but it has been found a close relationship between the relaxation process determined with DMA and the measured shape recovery process. In conclusion, the network structure properties give us an idea of the expected shape-fixation and retention of the temporary shape during the fixation and storage processes, while the network relaxation process is a powerful tool to define the shape-recovery process on unconstrained experiments. In addition, the Tprog has an important impact on the SMR, being a crucial parameter during loading stage and the shape-recovery process: not only the σb and εb can be adjusted by varying the Tprog , but also the shape-recovery process (enclosing the whole-process path) can be tuned as desired. Acknowledgments: The authors would like to thank Generalitat de Catalunya (2014-SGR-67) for its financial support.
MICINN
(MAT2014-53706-C03-01)
and
Author Contributions: Xavier Fernández-Francos and Silvia De la Flor conceived and designed the experiments, which were performed and analyzed by Alberto Belmonte. Dailyn Guzmán contributed on the samples preparation and characterization. Alberto Belmonte wrote the paper. Silvia De la Flor and Xavier Fernández-Francos revised the paper. Conflicts of Interest: The authors declare no conflict of interest. References 1. 2. 3. 4. 5.
Lendlein, A.; Sauter, T. Shape-memory effect in polymers. Macromol. Chem. Phys. 2013, 214, 1175–1177. [CrossRef] Julich-Gruner, K.K.; Löwenberg, C.; Neffe, A.T.; Behl, M.; Lendlein, A. Recent trends in the chemistry of shape-memory polymers. Macromol. Chem. Phys. 2013, 214, 527–536. [CrossRef] Santhosh Kumar, K.S.; Biju, R.; Reghunadhan Nair, C.P. Progress in shape memory epoxy resins. React. Funct. Polym. 2013, 73, 421–430. [CrossRef] Mather, P.T.; Luo, X.; Rousseau, I.A. Shape memory polymer research. Annu. Rev. Mater. Res. 2009, 39, 445–471. [CrossRef] Sun, L.; Huang, W.M.; Ding, Z.; Zhao, Y.; Wang, C.C.; Purnawali, H.; Tang, C. Stimulus-responsive shape memory materials: A review. Mater. Des. 2012, 33, 577–640. [CrossRef]
2162
Polymers 2015, 7, 2146–2164
6. 7. 8. 9. 10. 11. 12. 13. 14.
15. 16.
17.
18. 19. 20. 21. 22. 23.
24.
25. 26. 27. 28.
Rousseau, I.A.; Xie, T. Shape memory epoxy: Composition, structure, properties and shape memory performances. J. Mater. Chem. 2010, 20, 3431–3441. [CrossRef] Wei, K.; Zhu, G.; Tang, Y.; Niu, L. Shape-memory effects of a hydro-epoxy resin system. J. Polym. Res. 2013, 20, 123. [CrossRef] Feldkamp, D.M.; Rousseau, I.A. Effect of chemical composition on the deformability of shape-memory epoxies. Macromol. Mater. Eng. 2011, 296, 1128–1141. [CrossRef] Yakacki, C.M.; Willis, S.; Luders, C.; Gall, K. Deformation limits in shape-memory polymers. Adv. Eng. Mater. 2008, 10, 112–119. [CrossRef] Feldkamp, D.M.; Rousseau, I.A. Effect of the deformation temperature on the shape-memory behavior of epoxy networks. Macromol. Mater. Eng. 2010, 295, 726–734. [CrossRef] Zheng, N.; Fang, G.; Cao, Z.; Zhao, Q.; Xie, T. High strain epoxy shape memory polymer. Polym. Chem. 2015, 6, 3046–3053. [CrossRef] Arrieta, S.; Diani, J.; Gilormini, P. Experimental characterization and thermoviscoelastic modeling of strain and stress recoveries of an amorphous polymer network. Mech. Mater. 2014, 68, 95–103. [CrossRef] Arrieta, J.S.; Diani, J.; Gilormini, P. Cyclic and monotonic testing of free and constrained recovery properties of a chemically crosslinked acrylate. J. Appl. Polym. Sci. 2014. [CrossRef] Leonardi, A.B.; Fasce, L.A.; Zucchi, I.A.; Hoppe, C.E.; Soulé, E.R.; Pérez, C.J.; Williams, R.J.J. Shape memory epoxies based on networks with chemical and physical crosslinks. Eur. Polym. J. 2011, 47, 362–369. [CrossRef] Santiago, D.; Fernández-Francos, X.; Ferrando, F.; de la Flor, S. Shape-memory effect in hyperbranched poly(ethyleneimine)-modified epoxy thermosets. J. Polym. Sci. Polym. Phys. 2015, 924–933. [CrossRef] Yakacki, C.M.; Shandas, R.; Lanning, C.; Rech, B.; Eckstein, A.; Gall, K. Unconstrained recovery characterization of shape-memory polymer networks for cardiovascular applications. Biomaterials 2007, 28, 2255–2263. [CrossRef] [PubMed] Yakacki, C.M.; Shandas, R.; Safranski, D.; Ortega, A.M.; Sassaman, K.; Gall, K. Strong, tailored, biocompatible shape-memory polymer networks. Adv. Funct. Mater. 2008, 18, 2428–2435. [CrossRef] [PubMed] Chen, X.; Nguyen, T.D. Influence of thermoviscoelastic properties and loading conditions on the recovery performance of shape memory polymers. Mech. Mater. 2011, 43, 127–138. [CrossRef] Binder, W.H.; Sachsenhofer, R. “Click” chemistry in polymer and materials science. Macromol. Rapid Commun. 2007, 28, 15–54. [CrossRef] Moses, J.E.; Moorhouse, A.D. The growing applications of click chemistry. Chem. Soc. Rev. 2007, 36, 1249–1262. [CrossRef] [PubMed] Kolb, H.C.; Finn, M.G.; Sharpless, K.B. Click chemistry: Diverse chemical function from a few good reactions. Angew. Chem. Int. Ed. Engl. 2001, 40, 2004–2021. [CrossRef] Brändle, A.; Khan, A. Thiol–epoxy “click” polymerization: Efficient construction of reactive and functional polymers. Polym. Chem. 2012, 3, 3224–3227. [CrossRef] Flores, M.; Tomuta, A.M.; Fernández-Francos, X.; Ramis, X.; Sangermano, M.; Serra, A. A new two-stage curing system: Thiol-ene/epoxy homopolymerization using an allyl terminated hyperbranched polyester as reactive modifier. Polymer 2013, 54, 5473–5481. [CrossRef] Carlborg, C.F.; Vastesson, A.; Liu, Y.; van der Wijngaart, W.; Johansson, M.; Haraldsson, T. Functional off-stoichiometry thiol-ene-epoxy thermosets featuring temporally controlled curing stages via an UV/UV dual cure process. J. Polym. Sci. Polym. Chem. 2014, 52, 2604–2615. [CrossRef] Guzmán, D.; Ramis, X.; Fernández-Francos, X.; Serra, A. New catalysts for diglycidyl ether of bisphenol A curing based on thiol-epoxy click reaction. Eur. Polym. J. 2014, 59, 377–386. [CrossRef] Guzmán, D.; Ramis, X.; Fernández-Francos, X.; Serra, A. Enhancement in the glass transition temperature in latent thiol-epoxy click cured thermosets. Polymers 2015, 7, 680–694. [CrossRef] Nair, D.P.; Cramer, N.B.; Scott, T.F.; Bowman, C.N.; Shandas, R. Photopolymerized thiol-ene systems as shape memory polymers. Polymer 2010, 51, 4383–4389. [CrossRef] [PubMed] Sunitha, K.; Santhosh Kumar, K.S.; Mathew, D.; Reghunadhan Nair, C.P. Shape memory polymers (SMPs) derived from phenolic cross-linked epoxy resin via click chemistry. Mater. Lett. 2013, 99, 101–104. [CrossRef]
2163
Polymers 2015, 7, 2146–2164
29.
30. 31. 32. 33. 34. 35. 36. 37. 38.
39. 40. 41. 42.
McBride, M.K.; Gong, T.; Nair, D.P.; Bowman, C.N. Photo-mediated copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC) “click” reactions for forming polymer networks as shape memory materials. Polymer 2014, 55, 5880–5884. [CrossRef] [PubMed] Graessley, W.W. Statistical mechanics of random coil networks. Rubber Chem. Technol. 1975, 48, 1008–1017. [CrossRef] Miller, D.R.; Macosko, C.W. A new derivation of postgel properties of network polymers. Rubber Chem. Technol. 1976, 49, 1219–1231. [CrossRef] Charlesworth, J.M. Effect of crosslink density on molecular relaxations in diepoxide-diamine network polymers. Part 2. The rubbery plateau region. Polym. Eng. Sci. 1988, 28, 230–236. [CrossRef] Lesser, A.; Crawford, E. The role of network architecture on the glass transition temperature of epoxy resins. J. Appl. Polym. Sci. 1997, 66, 387–395. [CrossRef] Pascault, J.-P.; Sautereau, H.; Verdu, J.; Williams, R.J.J. Thermosetting Polymers; CRC Press: New York, NY, USA, 2002. Charlesworth, J.M. The glass transition temperature for non-gaussian network polymers. J. Macromol. Sci. 1987, 26, 105–133. [CrossRef] Hale, A.; Macosko, C.W.; Bair, H.E. Glass transition temperature as a function of conversion in thermosetting polymers. Macromolecules 1991, 24, 2610–2621. [CrossRef] Liu, Y.; Han, C.; Tan, H.; Du, X. Thermal, mechanical and shape memory properties of shape memory epoxy resin. Mater. Sci. Eng. A 2010, 527, 2510–2514. [CrossRef] Fan, M.; Yu, H.; Li, X.; Cheng, J.; Zhang, J. Thermomechanical and shape-memory properties of epoxy-based shape-memory polymer using diglycidyl ether of ethoxylated bisphenol-A. Smart Mater. Struct. 2013. [CrossRef] Rousseau, I.A.; Xie, T. Relationship between materials properties and shape memory behavior in epoxy-amine polymers. MRS Proc. 2009, 1190, 31–36. [CrossRef] Anthamatten, M.; Roddecha, S.; Li, J. Energy storage capacity of shape-memory polymers. Macromolecules 2013, 46, 4230–4234. [CrossRef] Pandini, S.; Bignotti, F.; Baldi, F.; Passera, S. Network architecture and shape memory behavior of cold-worked epoxies. J. Intell. Mater. Syst. Struct. 2013, 24, 1583–1597. [CrossRef] Gu, Y.; Li, S.X. Thermo-mechanical behavior of epoxy shape memory polymer. Adv. Mater. Res. 2013, 721, 169–172. [CrossRef] © 2015 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons by Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).
2164
